ES2710625T3 - Process of memory management of virtual machine, physical main machine, PCIE device and configuration procedure thereof, and migration management device - Google Patents

Abstract

A method for managing a memory of a virtual machine, in which the virtual machine is executed in a physical central computer, the physical central computer is connected to an express interface interconnection device, PCIE, the PCIE device is a device for hardware that supports the single-root I / O virtualization standard, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, in which a unique function number is configure for each PF and each VF separately; wherein the virtual machine executes a service using a VF of the at least one VF, and the virtual machine corresponds to a function number of the VF of the at least one VF; and the method comprises: assigning, by means of the virtual machine, a memory to at least one service located in the PCIE device, where the memory allocated to each service comprises multiple blocks of memory, each of the multiple blocks of memory being used to store operating information of a corresponding service, and each memory block is a section of memory space in which physical guest addresses are consecutive; generate, through the virtual machine, a table of base addresses, BAT, and a chip logical address table, CLAT, according to the memory assigned to each service, where the BAT comprises a base CLAT entry address corresponding to each service, and the CLAT comprises a first address of each memory block; send to the PCIE device, via the virtual machine, a BAT address and a function number corresponding to the virtual machine, where a correspondence between the BAT address and the function number corresponding to the virtual machine is stored in a table of virtual machine configuration, VCT, of the PCIE device connected to the physical central computer; wherein when the virtual machine is migrated from the physical central computer to a destination physical host, the memory blocks, the BAT and the CLAT are migrated to the migrated virtual machine in the destination physical host; and in which after migration, a correspondence between the address of the BAT and a function number corresponding to the migrated virtual machine, and is of a VF in a PCIE device connected to the destination physical host, is configured in a VCT on the PCIE device connected to the target physical host.

Description

DESCRIPTION

Process of memory management of virtual machine, physical main machine, PCIE device and configuration procedure thereof, and migration management device

Technical field

The present invention relates to the field of virtual machines and, in particular, to a method for managing a memory of a virtual machine, a physical central computer, a PCIE device and a configuration procedure thereof, and a device for migration management.

Background

In a typical application architecture of a virtual machine, a virtual machine manager (in English: Virtual Machine Manager, VMM for short) that provides a management function of a virtual machine is installed in a physical central computer. The VMM manages one or more virtual machines, and each virtual machine can run an operating system (in English: Operating System, OS for short) and various applications. In general, the hardware of the physical central computer mainly includes: certain hardware related to the execution of a virtual machine, for example, a central processing unit (in English: Central Processing Unit, CPU for short), a memory, a disk hard and certain hardware with relatively independent functions that is not directly configured to run a virtual machine, but is configured to run some particular services, for example, an peripheral interconnection device (PCIE) (such as a network adapter) and a video card) of the physical central computer.

The single-root I / O virtualization (in English: Single-Root I / O Virtualization, SR-IOV for short) is an input / output virtualization technology (in English: Input / Output, I / O for its acronym). Thanks to the technology, multiple virtual PCIE adapters can be obtained through virtualization in a physical PCIE adapter. This is known as virtual function (in English: Virtual Function, VF by its abbreviations). Using a network adapter that supports the SR-IOV as an example, multiple virtual network adapters can be obtained through virtualization on a physical network adapter, and one VF of each virtual machine corresponds to a virtual network adapter.

After being virtualized, a server can easily implement a live migration of a virtual machine. A live migration technology for a virtual machine refers to a virtual machine being transferred from a physical central computer to another physical central computer through a live migration, and the processing of the services is hardly affected in the process. of migration.

In an implementation process of the present invention, the inventor has observed that the prior art has at least the following problems:

In the prior art, during the live migration of a virtual machine, the data stored in a memory have to be migrated to a physical host computer, and the record states stored in a PCIE adapter also have to be migrated to the computer physical central destination. These registries are generally configured to register operation information of the services processed by the PCIE adapter. The operating information is used for the collection of service statistics, the control of services and the recording of service states. However, the registries that keep the operating information of the services are very decentralized. If a live migration is performed, the states of these records must be saved and restored individually, which is very cumbersome.

The document US 2012/0042034 A1 describes that before the migration of a virtual machine (VM), the VM directly accesses a storage unit using a virtual function (VF), in particular the direct VF access VF from the VM to the storage unit, therefore substantially without the intervention of a virtualization intermediary. During the migration of the VM, the status information is transferred from the VF to the virtualization intermediary, the VM accesses the storage unit using a physical function (PF) instead of the VF, and the virtualization intermediary intervenes in the IO access from the MV to the storage unit. Document US 2012/0137292 A1 discloses that a migration originating computer carries out a processing to send physical address information of the central computer of an MMIO region corresponding to a device used by a virtual machine through I / O addressed to a destination computer of migration. The migration destination computer performs processing to allocate an MMIO region corresponding to the physical address information of the central computer sent by the migration originating computer, to the device connected to the computers by means of the switching device.

Document WO 2014/032233 A1 discloses that the management node is used to convert a connection relationship between a VF module used by the M ^v before the live migration and a source server in a connection relationship between the VF module and a destination server through the PCIE switch.

Summary

To solve the problem of the prior art, the present invention provides a method for managing a memory of a virtual machine, a physical central computer, a PCIE device and a configuration procedure thereof, and a migration management device. The technical solutions are the following:

According to the independent claims there is provided a method for managing a memory of a virtual machine, a physical central computer, a PClE device and a configuration procedure thereof, and a migration management device. The dependent claims provide preferred embodiments.

According to a first aspect, an embodiment of the present invention provides a method for managing a memory of a virtual machine, where the virtual machine is executed in a physical central computer, the physical central computer is connected to an interconnection device. peripheral component expression, PCIE, the PCIE device is a hardware device that supports the single-root I / O virtualization standard, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, where a unique function number is configured for each PF and each VF separately; and the procedure includes:

assign, by means of the virtual machine, a memory to at least one service located in the PCIE device, where the memory assigned to each service includes multiple blocks of memory, each one of the multiple blocks of memory is used to store operating information of a corresponding service, and each memory block is a section of memory space in which the physical guest addresses are consecutive;

generate, through the virtual machine, a table of base addresses, BAT, and a table of logical addresses of chip, CLAT, according to the memory assigned to each service, where the BAT includes a base entry address of CLAT corresponding to each service, and the CLAT includes a first address of each block of memory; Y

send to the PCIE device, through the virtual machine, a BAT address and a function number corresponding to the virtual machine.

In a first possible way of implementing the first aspect, the at least one service includes a service to be processed, and the procedure also includes:

receive a request message that is used to obtain operation information of the service to be processed and that is sent by the PCIE device, where the request message includes the address of the BAT, a service type of the service to be processed and a number of service of the service to be processed; and sending, to the PCIE device according to the request message, operation information in a block of memory corresponding to the service to be processed.

According to the first possible way of implementing the first aspect, sending, to the PCIE device according to the request message, operating information in a memory block corresponding to the service to be processed includes:

obtain the BAT corresponding to the address of the BAT;

determine, according to the type of service of the service to be processed and the BAT, a base entry address of CLAT corresponding to the type of service;

determine an entry displacement according to the service number of the service to be processed; obtain, in accordance with the CLAT entry base address and the entry offset, a CLAT entry corresponding to the service to be processed; Y

send, to the PCIE device, operating information in a block of memory corresponding to the CLAT input corresponding to the service to be processed.

According to the second possible way of implementing the first aspect, in a third way of possible implementation of the first aspect, the CLAT is a single level CLAT or a multiple level CLAT; the CLAT of multiple levels includes N levels of CLAT, where N> 2, and N is an integer number; each level of CLAT includes multiple entries; each entry in the nth level of CLAT is used to indicate a direction of the level (n + 1) -this CLAT, where 2 <n <n + 1 <N, and n is an integer; and each entry in the N-th level of CLAT is used to indicate a first address of one of the memory blocks. According to the third possible way of implementing the first aspect, in a fourth way of possible implementation of the first aspect, the sizes of the multiple blocks of memory assigned to the service to be processed are the same; and determining an entry displacement according to the service number of the service to be processed includes:

when the CLAT is a CLAT of a single level, multiply the service number of the service to be processed by a size of a memory occupied by the operating information of the service to be processed, and divide the result of the multiplication by the size of the block of memory assigned to the service to be processed, to obtain a quotient and a remainder for the CLAT of a single level, where when the rest for the CLAT of a single level is greater than 0, the entry displacement is equal to the quotient for the CLAT of a single level, or when the remainder for the CLAT of a single level is equal to 0, the entry displacement is equal to the quotient for the CLAT of a single level minus 1; or

when the CLAT is a CLAT of multiple levels, where the entry displacement includes an entry displacement of each level of CLAT, calculate an entry displacement of the X-th level of CLAT in the following way:

where when a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the x-th level of CLAT is equal to 0, the entry displacement of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the operation information of the service to be processed, B is the size of the memory block assigned to the service to be processed, and C is the number of entries in the N-th level of CLAT.

According to a second aspect, an embodiment of the present invention further provides a method for configuring an peripheral interconnect device of peripheral components, PCIE, where the PCIE device is connected to a physical central computer, at least one virtual machine is running on the physical central computer, the PCIE device is a hardware device that supports the single-root I / O virtualization standard, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, where a unique function number is configured for each PF and each VF separately; each virtual machine includes a table of base addresses, BAT, where the BAT is generated by a virtual machine, to which the bAt belongs, according to a memory assigned to at least one service located in the PCIE device; the memory allocated to each service includes multiple blocks of memory, each of the multiple blocks of memory is used to store operating information of a corresponding service, each memory block is a section of memory space in which the physical addresses of Guest are consecutive; the BAT includes a base entry address of the logical chip address table, CLAT, corresponding to each service, and the CLAT includes a first address of each memory block; Y

The procedure includes:

receive, through the PCIE device, a address from the base address table, BAT, in each virtual machine and a function number corresponding to each virtual machine; Y

save a correspondence between the address of the BAT and the function number in a virtual machine configuration table, VCT.

In a first possible way of implementing the second aspect, the at least one service includes a service to be processed, and the procedure also includes:

determine, according to the service information of the service to be processed and the VCT, a BAT address corresponding to the service to be processed; Y

obtain, from a memory corresponding to the service to be processed, operating information of the service to be processed in accordance with the BAT address corresponding to the service to be processed, a service type of the service to be processed and a service number of the service to be processed. process.

According to the first possible way of implementing the second aspect, in a second way of possible implementation of the second aspect, determining, according to the service information of the service to be processed and the VCT, a BAT address corresponding to the service to be processed includes :

receive the service information, sent by one side of a central computer or a network side, of the service to be processed;

determine a function number corresponding to the service to be processed, in accordance with a feature of the service information, sent by the central computer side, of the service to be processed or in accordance with a PCIE interface used when the service information is received, transmitted by the central computer side of the service to be processed, where the feature is used to indicate the function number corresponding to the service to be processed; Y

determine, in accordance with the VCT and the function number determined corresponding to the service to be processed, the address of the BAT corresponding to the function number corresponding to the service to be processed.

According to the second way of possible implementation of the second aspect, in a third way of possible implementation of the second aspect, obtaining, from a memory corresponding to the service to be processed, operation information of the service to be processed according to the address of the BAT corresponding to the service to be processed, a service type of the service to be processed and a service number of the service to be processed includes:

send a request message to a virtual machine corresponding to the function number determined that corresponds to the service to be processed, where the request message includes the address of the BAT corresponding to the function number corresponding to the service to be processed, the type of service of the service to be processed and the service number of the service to be processed; Y

receive operating information of the service to be processed, sent by the virtual machine corresponding to the function number corresponding to the service to be processed, where the operating information of the service to be processed is obtained from a block of memory corresponding to the message of request by means of the virtual machine corresponding to the number of function that corresponds with the service to be processed.

According to the first or second way of possible implementation of the second aspect, in a fourth way of possible implementation of the second aspect, obtaining, from a memory corresponding to the service to process, operation information of the service to be processed according to with the address of the BAT corresponding to the service to be processed, a service type of the service to be processed and a service number of the service to be processed includes:

read, according to the address of the BAT corresponding to the service to be processed, the BAT corresponding to the service to be processed;

determine, according to the type of service of the service to be processed, a base CLAT entry address corresponding to the type of service of the service to be processed of the BAT corresponding to the service to be processed; determine an entry displacement according to the service number of the service to be processed; read a CLAT entry corresponding to the entry displacement and the CLAT entry base address that corresponds to the service type of the service to be processed; Y

read operation information in a block of memory corresponding to the CLAT input.

According to the fourth possible way of implementing the second aspect, in a fifth way of possible implementation of the second aspect, the CLAT is a single level CLAT or a multiple level CLAT; the CLAT of multiple levels includes N levels of CLAT, where N> 2, and N is an integer number; each level of CLAT includes multiple entries; each entry in the nth level of CLAT is used to indicate a direction of the level (n + 1) -this CLAT, where 2 <n <n + 1 <N, and n is an integer; and each entry in the N-th level of CLAT is used to indicate a first address of one of the memory blocks.

According to the fifth possible way of implementation of the second aspect, in a sixth possible way of implementing the second aspect, the sizes of the multiple blocks of memory assigned to the service to be processed are the same; Y

determining an entry displacement according to the service number of the service to be processed includes: when the CLAT is a single-level CLAT, multiply the service number of the service to be processed by a size of a memory occupied by the information of operation of the service to be processed, and divide the result of the multiplication by the size of the memory block assigned to the service to be processed, to obtain a quotient and a remainder for the CLAT of a single level, where when the rest for the CLAT of a only level is greater than 0, the entry displacement is equal to the quotient for the CLAT of a single level, or when the rest for the CLAT of a single level is equal to 0, the entry displacement is equal to the quotient for the CLAT of a single level minus 1; or

when the CLAT is a CLAT of multiple levels, where the entry displacement includes an entry displacement of each level of CLAT, calculate an entry displacement of the X-th level of CLAT in the following way:

where when a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the level If CLAT is equal to 0, the input displacement of the CLAT x-th level is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the information of operation of the service to be processed, B is the size of the memory block assigned to the service to be processed, and C is the number of entries in the N-th level of CLAT.

According to a third aspect, an embodiment of the present invention also provides a method for managing a memory of a virtual machine, wherein the method includes:

obtaining, by means of a migration management device in a physical host computer, a address of a table of base addresses, BAT, in a virtual machine of a physical home computer during a live migration, where the physical central computer of destination and the originating physical host computer are connected to different peripheral interconnect devices of peripheral components, PCIE, the PCIE device is a hardware device that supports the single-root I / O virtualization standard, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, where a unique function number is configured for each PF and each VF separately; obtain a function number corresponding to a migrated virtual machine; Y

configure a correspondence between the address of the BAT and the function number in a virtual machine configuration table, VCT, in a PCIE device connected to the physical host computer.

According to a fourth aspect, an embodiment of the present invention also provides a physical central computer, where a virtual machine is executed in the physical central computer, the physical central computer is connected to an express interconnection device of peripheral components, PCIE, the PCIE device is a hardware device that supports the single-root I / O virtualization standard, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, where a unique function number is configured for each PF and each VF separately; and the virtual machine includes:

an assignment module, configured to assign a memory to at least one service located in the PCIE device, where the memory assigned to each service includes multiple blocks of memory, each of the multiple blocks of memory is used to store operating information of a corresponding service, and each memory block is a section of memory space in which the physical guest addresses are consecutive;

a module of generation of inputs, configured to generate a table of base addresses, BAT, and a table of logical addresses of chip, CLAT, according to the memory assigned to each service, where the BAT includes a base entry address of CLAT corresponding to each service, and the CLAT includes a first address of each memory block; Y

a sending module, configured to send to the PCIE device a BAT address and a function number corresponding to the virtual machine.

In a first possible way of implementing the fourth aspect, the at least one service includes a service to be processed, and the virtual machine also includes:

a receiving module, configured to receive a request message that is used to obtain operation information of the service to be processed and sent by the PCIE device, where the request message includes the address of the BAT, a service type of the service to be processed and a service number of the service to be processed; Y

the sending module is also configured to send, to the PCIE device according to the request message, operation information in a memory block corresponding to the service to be processed. According to the first possible way of implementing the fourth aspect, in a second way of possible implementation of the fourth aspect, the sending module is specifically configured to:

obtain the BAT corresponding to the address of the BAT; determine, according to the type of service of the service to be processed and the BAT, a base entry address of CLAT corresponding to the type of service; determine an entry displacement according to the service number of the service to be processed; obtain, in accordance with the CLAT entry base address and the entry offset, a CLAT entry corresponding to the service to be processed; and sending, to the PCIE device, operating information in a block of memory corresponding to the CLAT entry corresponding to the service to be processed.

According to the second possible way of implementing the fourth aspect, in a third possible way of implementing the fourth aspect, the CLAT is a single level CLAT or a multiple level CLAT; the CLAT of multiple levels includes N levels of CLAT, where N> 2, and N is an integer number; each level of CLAT includes multiple entries; each entry in the nth level of CLAT is used to indicate a direction of the level (n + l) -this CLAT, where 2 <n <n + 1 <N, and n is an integer; and each entry in the N-th level of CLAT is used to indicate a first address of one of the memory blocks.

According to the third possible way of implementing the fourth aspect, in a fourth possible way of implementing the fourth aspect, the sizes of the multiple blocks of memory assigned to the service to be processed are the same; Y

the shipping module is specifically configured to:

when the CLAT is a CLAT of a single level, multiply the service number of the service to be processed by a size of a memory occupied by the operating information of the service to be processed, and divide the result of the multiplication by the size of the block of memory assigned to the service to process, to obtain a quotient and a remainder for the CLAT of a single level, where when the rest for the CLAT of a single level is greater than 0, the entry displacement is equal to the quotient for the CLAT of a single level, or when the rest for the CLAT of a single level is equal to 0, the entry displacement is equal to the quotient for the CLAT of a single level minus 1; or

when the CLAT is a CLAT of multiple levels, where the entry displacement includes an entry displacement of each level of CLAT, calculate an entry displacement of the X-th level of CLAT in the following way:

where when a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the x-th level of CLAT is equal to 0, the entry displacement of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the operation information of the service to be processed, B is the size of the memory block assigned to the service to be processed and C is the number of entries in the N-th level of CLAT.

According to a fifth aspect, an embodiment of the present invention further provides a PCIE device, where the PCIE device is connected to a physical central computer, at least one virtual machine is executed in the physical central computer, the PCIE device is a hardware device that satisfies the single-root I / O virtualization standard, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, where a unique function number it is configured for each PF and each VF separately; each virtual machine includes a table of base addresses, BAT, where the BAT is generated by a virtual machine, to which the BAT belongs, according to a memory assigned to at least one service located in the PCIE device; the memory allocated to each service includes multiple blocks of memory, each of the multiple blocks of memory is used to store operating information of a corresponding service, each memory block is a section of memory space in which the physical addresses of Guest are consecutive; the BAT includes a base entry address of the logical chip address table, CLAT, corresponding to each service, and the CLAT includes a first address of each memory block; and the PCIE device also includes:

a reception module, configured to receive a address from the base address table, BAT, in each virtual machine and a function number corresponding to each virtual machine; Y

a generation module, configured to store a correspondence between the BAT address and the function number in a virtual machine configuration table, VCT.

In a first possible way of implementing the fifth aspect, the at least one service includes a service to be processed, and the PCIE device also includes:

a determination module, configured to determine, in accordance with the service information of the service to be processed and the VCT, a address of a BAT corresponding to the service to be processed; Y

a obtaining module, configured to obtain, from a memory corresponding to the service to be processed, operation information of the service to be processed according to the BAT address corresponding to the service to be processed, a service type of the service to be processed and a service number of the service to be processed.

According to the first possible way of implementation of the fifth aspect, in a second way of possible implementation of the fifth aspect,

the reception module is further configured to receive the service information sent by one side of the central computer or a network side of the service to be processed; Y

The determination module is specifically configured to:

determine a function number corresponding to the service to be processed, in accordance with a feature of the service information, sent by the central computer side, of the service to be processed or in accordance with a PCIE interface used when the service information is received, transmitted by the central computer side of the service to be processed, where the feature is used to indicate the function number corresponding to the service to be processed; Y

determine, in accordance with the VCT and the function number determined corresponding to the service to be processed, the address of the BAT corresponding to the function number corresponding to the service to be processed.

According to the second possible way of implementing the fifth aspect, in a third possible way of implementing the fifth aspect, the obtaining module is specifically configured to:

send a request message to a virtual machine corresponding to the function number determined that corresponds to the service to be processed, where the request message includes the address of the BAT corresponding to the function number corresponding to the service to be processed, the type of service of the service to be processed and the service number of the service to be processed; and receive operating information of the service to be processed, sent by the virtual machine corresponding to the function number corresponding to the service to be processed, where the operation information of the service to be processed is obtained from a memory block corresponding to the request message through the virtual machine corresponding to the function number that corresponds to the service to be processed.

According to the first or second possible way of implementation of the fifth aspect, in a fourth possible implementation manner of the fifth aspect, the obtaining module is specifically configured to:

read, according to the address of the BAT corresponding to the service to be processed, the BAT corresponding to the service to be processed; determine, according to the type of service of the service to be processed, a base CLAT entry address corresponding to the type of service of the service to be processed of the BAT corresponding to the service to be processed; determine an entry displacement according to the service number of the service to be processed; read a CLAT entry corresponding to the entry displacement and the CLAT entry base address that corresponds to the service type of the service to be processed; and read operation information in a block of memory corresponding to the CLAT input.

According to the fourth possible implementation manner of the fifth aspect, in a fifth possible implementation manner of the fifth aspect, the CLAT is a single level CLAT or a multiple level CLAT; the CLAT of multiple levels includes N levels of CLAT, where N> 2, and N is an integer number; each level of CLAT includes multiple entries; each entry in the nth level of CLAT is used to indicate a direction of the level (n + l) -this CLAT, where 2 <n <n + 1 <N, and n is an integer; and each entry in the N-th level of CLAT is used to indicate a first address of one of the memory blocks.

According to the fifth possible way of implementation of the fifth aspect, in a sixth possible way of implementing the fifth aspect, the sizes of the multiple blocks of memory assigned to the service to be processed are the same; Y

the obtaining module is specifically configured to:

when the CLAT is a CLAT of a single level, multiply the service number of the service to be processed by a size of a memory occupied by the operating information of the service to be processed, and divide the result of the multiplication by the size of the block of memory assigned to the service to process, to obtain a quotient and a remainder for the CLAT of a single level, where when the rest for the CLAT of a single level is greater than 0, the entry displacement is equal to the quotient for the CLAT of a single level, or when the rest for the CLAT of a single level is equal to 0, the entry displacement is equal to the quotient for the CLAT of a single level minus 1; or

when the CLAT is a CLAT of multiple levels, where the entry displacement includes an entry displacement of each level of CLAT, calculate an entry displacement of the X-th level of CLAT in the following way:

two where when a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the x-th level of CLAT is equal to 0, the entry displacement of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the operation information of the service to be processed, B is the size of the memory block assigned to the service to be processed, and C is the number of entries in the N-th level of CLAT.

According to a sixth aspect, an embodiment of the present invention further provides a migration management device, wherein the migration management device is arranged in a physical host computer, and the migration management device includes:

a first module of obtaining, configured to obtain a direction of a table of base addresses, BAT, in a virtual machine of a physical central computer of origin during a live migration, where the physical central computer of destination and the physical central computer of are connected to different peripheral interconnection devices of peripheral components, PCIE, the PCIE device is a hardware device that supports the single-root I / O virtualization standard, SR-IOV, and the PCIE device has at least one function. physical, PF, and at least one virtual function, VF, where a unique function number is configured for each PF and each VF separately;

a second obtaining module, configured to obtain a function number corresponding to a migrated virtual machine; Y

a configuration module, configured to configure a correspondence between the address of the BAT and the function number in a virtual machine configuration table, VCT, in a PCIE device connected to the physical host computer.

According to a seventh aspect, one embodiment of the present invention further provides a physical central computer, wherein the physical central computer includes a processor, a memory, an input / output unit and a bus; the memory is configured to store an executable instruction by computer; the processor is connected to the memory using the bus; and when the physical central computer is in operation, the processor executes the executable instruction by computer stored in the memory, so that the computer executes the procedure for managing a memory of a virtual machine provided in the first aspect or the third aspect .

Beneficial effects of the technical solutions provided in the present invention are the following:

During service processing, operation information is sought in a memory block according to a feature in a service packet and a service number, and the operation information is read by means of a temporary storage unit. This is easy to do and requires less time. During the live migration of a virtual machine the problems caused by the migration of records are avoided, and a VMM automatically and directly migrates a block of memory, a BAT and a CLAT to a virtual machine of destination, provided that a correspondence between a number of function corresponding to a virtual machine migrated and a first address of the BAT in a virtual machine of a physical central computer of origin this configured in a VCT in a PCIE adapter after the migration. This is simple and easy to implement, greatly increases the speed of live migration of a virtual machine and reduces downtime.

Brief description of the drawings

To more clearly describe the technical solutions of the embodiments of the present invention, the attached drawings required to describe the embodiments are briefly set forth below. Obviously, the accompanying drawings of the following description show only some embodiments of the present invention, and other drawings may be obtained by those skilled in the art from these accompanying drawings without further investigation.

FIG. 1 is a diagram of an application scenario according to the embodiments of the present invention.

FIG. 2 is a flow chart of a method for managing a memory of a virtual machine according to the embodiment 1 of the present invention.

FIG. 3 is a flow chart of a method for configuring a PCIE device according to embodiment 2 of the present invention.

FIG. 4 is a flowchart of a method for managing a memory of a virtual machine according to the embodiment 3 of the present invention.

FIG. 5 is a flowchart of a method for managing a memory of a virtual machine of according to embodiment 4 of the present invention.

FIG. 6 is a schematic diagram of a single level CLAT according to the embodiment 4 of the present invention.

FIG. 7 is a schematic diagram of a multi-level CLAT according to the embodiment 4 of the present invention.

FIG. 8 is a structural block diagram of a physical central computer according to embodiment 5 of the present invention.

FIG. 9 is a structural block diagram of a physical central computer according to embodiment 6 of the present invention.

FIG. 10 is a structural block diagram of a PCIE device according to embodiment 7 of the present invention.

FIG. 11 is a structural block diagram of a PCIE device according to embodiment 8 of the present invention.

FIG. 12 is a structural block diagram of a migration management device according to embodiment 9 of the present invention.

Description of forms of realization

In order to understand more clearly the objectives, the technical solutions and the advantages of the present invention, the embodiments of the present invention are described in detail below with reference to the attached drawings.

To facilitate the description of the forms of embodiment, a scenario of application of the embodiments of the present invention is briefly described first. With reference to FIG. 1, a virtual machine manager, VMM, and at least one virtual machine are running a physical central computer. The VMM can manage one or more virtual machines. Each virtual machine can run an operating system, OS, and several applications. The physical central computer can be connected to a PCIE device, and the PCIE device is a hardware device that supports the single-root I / O virtualization standard, SR-IOv. The PCIE device has at least one physical function (in English: Physical Function, PF for short) and at least one virtual function, VF. A unique function number is configured for each PF and each VF separately. The virtual machine executes a particular service using a PF or a VF. Each PF or VF corresponds to a virtual machine. A correspondence between a function number of a PF or a function number of a VF and a virtual machine can be saved in the PCIE device.

The physical central computer includes hardware related to the execution of a virtual machine, for example, a CPU and at least one storage unit (a hard disk, a memory and the like) of a computer readable storage medium. In addition, the physical central computer may include a communications unit, an input unit, a display unit and the like. The PCIE device can be a PCIE adapter, for example, a network adapter and a video card.

The physical central computer may also include a migration management device. During the live migration of a virtual machine, a migration management device in a physical home computer migrates the data stored in a memory to a physical host of destination. In one embodiment, the migration management device may be the previous VMM.

It should be noted that the types of device and the above connection modes are only examples, and the present invention does not impose any limitations on them.

Form of realization 1

This embodiment of the present invention provides a method for managing a memory of a virtual machine. The procedure is executed by one of the at least one previous virtual machine. With reference to FIG. 2, the procedure includes the following stages.

Step 101. The virtual machine allocates a memory to at least one service located in a PCIE device, where the memory allocated to each service includes multiple blocks of memory, each of the multiple blocks of memory being used to store operating information of a corresponding service, and each memory block is a section of memory space in which the physical guest addresses are consecutive.

Stage 102. The virtual machine generates a table of base addresses (in English: Base Address Table, BAT by its acronyms) and a table of logical addresses of chip (in English: Chip Logic Address Table, CLAT by its abbreviations) in agreement with the memory assigned to each service, where the BAT includes a base entry address of CLAT corresponding to each service, and the CLAT includes a first address of each memory block.

Step 103. The virtual machine sends to the PCIE device a BAT address and a function number corresponding to the virtual machine, where the PCIE device is configured to: register a correspondence between the BAT address and the function number, and obtain, from the virtual machine, operating information of a service according to the correspondence between the BAT address and the function number.

The correspondence between the address of the BAT and the function number can be saved in a virtual machine configuration table (in English: VM Configuration Table, VCT for short).

In this embodiment of the present invention, a virtual machine assigns a memory to at least one service located in a PCIE device, generates a BAT and a CLAT according to the memory assigned to each service, and sends a BAT address. and a function number to the PCIE device, so that the PCIE device can register a correspondence between the BAT address and the function number. During the processing of services, the PCIE device obtains, from the virtual machine, operating information of a service according to the correspondence between the BAT address and the function number. This is easy to do and requires less time. During the live migration of a virtual machine the problems caused by the migration of records are avoided, provided that a block of memory, a BAT and a CLAT are migrated directly to a physical machine of destination, and that a correspondence between a number of function corresponding to a migrated virtual machine and a BAT address in a virtual machine of a physical central computer of this origin configured in a VCT in a PCIE device after the migration. This is simple and easy to implement, greatly increases the speed of live migration of a virtual machine and reduces downtime. Form 2

This embodiment of the present invention provides a method for configuring a PCIE device. The procedure is executed by the previous PCIE device. With reference to FIG. 3, the procedure includes the following stages.

Stage 201. The PCIE device receives a BAT address in each virtual machine and a function number corresponding to each virtual machine, where each virtual machine includes a table of base addresses, BAT, the BAT is generated by a virtual machine, a the one that ^b A ^t belongs to, according to a memory assigned to at least one service located in the PCIE device; the memory allocated to each service includes multiple blocks of memory, each of the multiple blocks of memory is used to store operating information of a corresponding service, each memory block is a section of memory space in which the physical addresses of Guest are consecutive; the BAT includes a base entry address of CLAT corresponding to each service, and a CLAT includes a first address of each memory block.

Step 202. Save a correspondence between the address of the BAT and the function number in a virtual machine configuration table, VCT, so that the VCT includes the correspondence between the function number and the BAT address.

In this embodiment of the present invention, a PCIE device receives a BAT address and a function number, and stores a correspondence between the address of the BAT and the function number in a virtual machine configuration table, ^v C ^t . During the processing of services, the PCIE device obtains, from a virtual machine, operating information of a service according to the correspondence between the BAT address and the function number. This is easy to do and requires less time. During the live migration of a virtual machine the problems caused by the migration of records are avoided, provided that a block of memory, a BAT and a CLAT are migrated directly to a physical host computer of destination, and that a correspondence between a number of function corresponding to a virtual machine migrated and a BAT address in a virtual machine of a physical central computer of origin this configured in a VCT in a PCIE device after the migration. This is simple and easy to implement, greatly increases the speed of live migration of a virtual machine and reduces downtime.

Form 3

This embodiment of the present invention provides a method for managing a memory of a virtual machine. The procedure is executed by the previous migration management device. With reference to FIG. 4, the procedure includes the following stages.

Step 301. The migration management device in a destination physical host obtains a address from a table of base addresses, BAT, in a virtual machine of a physical home computer during a Live migration, where the physical host computer of origin and the physical home computer are connected to different devices of peripheral interconnection of peripheral components, PCIE, the PCIE device is a hardware device that supports the I / O virtualization standard of a single root, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, where a unique function number is configured for each PF and each VF separately.

The migration management device in this embodiment may be a VMM.

Step 302. The migration management device obtains a function number corresponding to a migrated virtual machine.

Step 303. The migration management device configures a correspondence between the address of the BAT and the function number in a virtual machine configuration table, VCT, in a PCIE device connected to the physical host computer of destination.

In this embodiment of the present invention, during the live migration of a virtual machine the problems caused by the migration of records are avoided, provided that a block of memory, a BAT and a CLAT are automatically and directly migrated to a physical machine of destination during a migration process, and that a correspondence between a number of function corresponding to a virtual machine migrated and a BAT address in a virtual machine of a physical central computer of origin this configured in a VCT in a PCIE device after migration. This is simple and easy to implement, greatly increases the speed of live migration of a virtual machine and reduces downtime.

Form of realization 4

This embodiment of the present invention provides a method for managing a memory of a virtual machine. With reference to FIG. 5, the procedure includes the following stages.

Step 401. A virtual machine allocates a memory to at least one service located in a PCIE device, where the memory allocated to each service includes multiple blocks of memory, each of the multiple blocks of memory being used to store operating information of a corresponding service, and each memory block is a section of memory space in which the physical guest addresses (in English: Guest Physical Address, GPA for short) are consecutive.

When the virtual machine is started, the virtual machine is applied to a memory of a physical central computer for the PCIE device and allocates the memory that is applied to a service. Specifically, after receiving an application from the virtual machine, a PCIE device driver of the physical central computer allocates a memory to the virtual machine. The PCIE device driver of the physical central computer refers to a driver that is used to operate the PCIE device and is installed on a hard disk of the physical central computer.

In this embodiment, the operation information of the service can be configuration information, a context table, a service resource or a table about the environment. In the context table, various states of the service are recorded, such as an execution status, a queuing state, a timer status and a statistics collection state.

In this embodiment, the memory blocks assigned by the virtual machine to each service are of a constant size and include consecutive GPAs, that is, the sizes of the multiple memory blocks assigned to each service are identical. Specifically, a service can occupy independently a block of memory, or multiple services can occupy a block of memory together. In addition, in this embodiment, an unimportant size of the memory block is a size of a memory page and is generally 4K.

Step 402. The virtual machine generates a table of base addresses, BAT, and a table of logical addresses of chip, CLAT, according to the memory assigned to each service, where the BAT includes a base entry address of CLAT corresponding to each service, and the CLAT includes a first address of each memory block.

If there are several services of the same type of service in the CLAT, to differentiate between the different services of the same type of service, a service number can be assigned to each service. The service number can be used for the calculation of a CLAT input offset, and the details are described below. In this document, the type of service is a type of a service that can be executed by the PCIE device, for example, a divert service, a segmentation service and a checksum service, which is executed by an adapter of network.

In addition, when the memory is allocated to the service in step 401, a memory block may be further allocated. to BAT and CLAT. A number of bits in a first address of the memory block is constant, and therefore, a number of octets in the first address of the memory block is also constant. For example, the number of bits in the first address of the memory block is 64, and a number of octets occupied by a first address of each memory block is 8B. Therefore, if a size of a memory assigned to the CLAT is 4K, a maximum number of entries in each CLAT is 4K / 8 = 512. When the entries in a CLAT are not sufficient, it is necessary to generate a CLAT of multiple levels to store a first address of a block of memory. That is, the CLAT can be a single level CLAT or a multiple level CLAT. The CLAT of multiple levels includes N levels of CLAT, where N> 2 and N is an integer number. Each level of CLAT includes multiple entries. Each entry in the nth level of CLAT is used to indicate a direction of the level (n + l) -this CLAT, where 2 <n <n + 1 <N and n is an integer. Each entry of the N-th level of CLAT is used to indicate a first address of one of the memory blocks.

As shown in FIG. 6, in the CLAT of a single level, each entry of the CLAT corresponds to a block of memory.

As shown in FIG. 7, in the CLAT of multiple levels, which uses a CLAT of two levels as an example, an entry in the first level of CLAT corresponds to multiple entries in the second level of CLAT, and an entry in the second level of CLAT corresponds to a memory block.

Therefore, the BAT may further include a size of a CLAT entry (i.e., a size of a memory block allocated to the CLAT), the size of the memory block, a number of levels of the CLAT and the like. Step 403. The virtual machine sends to the PCIE device a BAT address and a function number corresponding to the virtual machine.

The address of the BAT can be a first address of the BAT. The function number is a function number of a PF or VF corresponding to the virtual machine.

Specifically, the function number of the PF or VF corresponding to the virtual machine can be specified by the virtual machine or by a VMM. Therefore, step 403 may be performed, alternatively, by the VMM. Furthermore, in this embodiment, the virtual machine can also store a correspondence between the BAT address and the function number for later use.

Step 404. The PCIE device receives a BAT address in each virtual machine and a function number corresponding to each virtual machine, and it keeps a correspondence between the BAT address and the function number in a virtual machine configuration table , VCT, so that the VCT includes the correspondence between the function number and the address of the BAT.

The VCT is stored in the PCIE device (for example, a network adapter), and the BAT and CLAT are saved in the virtual machine. Specifically, the VCT can be generated in advance, and after receiving the BAT address and the function number, the PCIE device writes the address of the BAT and the function number in the VCT. More specifically, the VCT can be generated by a VMM or the PCIE.

Specifically, the address of the BAT can be a physical address of the physical central computer (also called the physical address of the central computer, in English: Host Physical Address, or HPA by its initials), or a physical address of the virtual machine (also called GPA). ). If the address of the BAT is the GPA, when the address is accessed, the PCIE device needs to implement a conversion from GPA to HPA through the use of VT-D (an Intel technology) or IOMMU (an AMD technology) in a CPU, or other technology that can automatically implement the conversion of addresses from GPA to HPA.

Step 405. The PCIE device determines, according to the service information of a service to be processed and the VCT, a address of a BAT corresponding to the service to be processed, where the at least one service includes the service to be processed.

The service information of the service to be processed includes service information sent by a central computer side or service information sent by a network side. The service information sent by the central computer side is generally service data. The service information sent by the network side is generally a service pack.

Specifically, step 405 can be implemented in the following manner:

receive the service information sent by the central computer side or the network side of the service to be processed;

determine a function number corresponding to the service to be processed, in accordance with a feature of the service information sent by the central computer side of the service to be processed or agreed with a PCIE interface used when receiving the service information, transmitted by the central computer side, of the service to be processed, where the feature is used to indicate the function number corresponding to the service to be processed; Y

determine, in accordance with the VCT and the function number determined corresponding to the service to be processed, the address of the BAT corresponding to the function number corresponding to the service to be processed.

In this embodiment, a feature in the service packet sent by the network side may be a field in the service pack. A field used as a feature in this document may vary with a type of a PCIE device. For example, when the PCIE device is a network adapter, a destination access control address (MAC) address can be used as a feature. That is, when a service packet is received, the network adapter can determine a corresponding function number according to the destination MAC address in the service pack. In a case in which the central computer side sends the service data, a corresponding function number can be determined according to a PCIE interface used when the service data transmitted by the central computer side is received.

The central computer side refers to a system that includes a CPU, a memory and a hard disk that are from a physical central computer. The network side refers to an external network side connected to the physical central computer. Step 406. The PCIE device obtains, from a memory corresponding to the service to be processed, operating information of the service to be processed in accordance with the BAT address corresponding to the service to be processed, a service type of the service to be processed and a service number of the service to be processed.

In this embodiment, step 406 can be implemented in the following two ways.

Way of implementation 1:

In the first place, the PCIE device sends a request message to a virtual machine corresponding to the determined function number that corresponds to the service to be processed. The request message includes the address of the BAT corresponding to the function number corresponding to the service to be processed, the type of service of the service to be processed and the service number of the service to be processed. The type of service is a type of a service corresponding to the service pack or service data received by the PCIE device. The service corresponding to the service package is a service located in the service package. The service corresponding to the service data is a service transported in a service package that is obtained after the PCIE device processes the service data.

Secondly, the virtual machine receives the request message that is used to obtain the operation information of the service to be processed and that is sent by the PCIE device.

Third, the virtual machine sends, to the PCIE device according to the request message, operation information in a block of memory corresponding to the service to be processed.

Specifically, this stage can include the following stages.

Stage 1. The virtual machine obtains the BAT corresponding to the address of the BAT.

In addition, the request message can alternately carry only the function number but not include the address of the BAT. The virtual machine obtains the BAT according to the function number and correspondence, stored in the virtual machine, between the BAT address and the function number.

Stage 2. The virtual machine determines, according to the type of service of the service to be processed and the BAT, a base entry address of CLAT corresponding to the type of service.

Stage 3. The virtual machine determines an entry displacement according to the service number of the service to be processed.

Specifically, the sizes of multiple blocks of memory assigned to the service to be processed are identical. The virtual machine calculates the input displacement in the following way: When the CLAT is a single-level CLAT, the service number of the service to be processed is multiplied by a size of a memory occupied by the operating information of the service to be processed. , and the result of the multiplication is divided by the size of the memory block assigned to the service to be processed to obtain a quotient and a remainder for the CLAT of a single level. When the rest for the CLAT of a single level is greater than 0, the entry displacement is equal to the quotient for the CLAT of a single level, or when the rest for the CLAT of a single level is equal to 0, the displacement of entry is equal to the quotient for the CLAT of a single level minus 1.

When the CLAT is a multi-level CLAT, where the input offset includes an entry offset of each CLAT level, an entry offset of the CLAT X-level is calculated as follows:

When a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the x-th level of CLAT is equal to 0, the entry displacement of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the operation information of the service to be processed, B is the size of the memory block assigned to the service to be processed and C is the number of entries in the N-th level of CLAT.

When the input displacement of the CLAT x-th level is 0, it indicates that a corresponding entry is the first entry in the x-th level of CLAT, when the entry offset of the x-th level of CLAT is 1, it indicates that a corresponding entry is the second entry in the x-th level of CLAT, and so on. It should be noted that a remainder for the N-th level of CLAT (ie, the last CLAT level) is used to indicate the operation information, corresponding to the service number, in a given memory block. When the remainder for the Nth level of CLAT is equal to 0, the operation information corresponding to the service number is the last element of operation information in the given memory block. When the remainder for the N-th level of CLAT is greater than 0, the operation information corresponding to the service number is the m-th element of operating information in the given memory block, and m is equal to the rest for the level N-th of CLAT.

The calculation of the input offset is described below by means of an example. For example, a size of each used memory is 1 KB, a size of a memory block is 4 KB, a number of bits in a first address of a memory block is 64, and a size of a block of memory is 64. memory assigned to a CLAT is also 4 KB.

As regards a single level CLAT, it is assumed that there are 2K services in the virtual machine, and 512 memory blocks with a size of 4K are assigned to the 2K services in the 401 stage. First addresses of the 512 blocks of memory are stored in a CLAT. When a service number in the received request message is 100, an entry shift is calculated as follows: 100 * 1K / 4K = 25, where the quotient is 25 and the remainder is 0. Therefore, the The input offset is 24, and the operation information corresponding to a service whose service number is 100 is the last operating information element in a memory block corresponding to the 25th input.

With respect to a multi-level CLAT, it is assumed that there are 16K services in the virtual machine, and a two-tier CLAT can support 512 * 512 * 4 = 1M services. Therefore, if it is necessary to support the 16K services and the two-tier CLAT is used, the first CLAT level only has to include eight entries. When the virtual machine is started, 4K memory blocks with a size of 4K are assigned to the 16K services. Then, a block of memory with the size of 4K is used to generate the first level of CLAT, and eight blocks of memory with the size of 4K are used to generate the second level of CLAT.

When a service number in the received request message is 2054, an entry offset of the first CLAT level is calculated as follows: 2054 * 1024/4096 / 5122'1 = 1, where the remainder is 6. therefore, the entry displacement of the first level of CLAT is 1, that is, it is determined that an entry in the first level of CLAT is the second entry. An input offset of the second CLAT level is calculated as follows: 6 * 1024/4096 / 5122'2 = 1, where the remainder is 2. Therefore, the input offset of the second CLAT level is 1, and the operating information corresponding to a service whose service number is 2054 is the second operating information element in a memory block corresponding to the second entry in the second level of CLAT.

The service number is assigned when each service starts. For example, in a transmission control protocol (TCP) service, a service number is assigned to the service when a TCP connection is established. The service number can be assigned sequentially from 1 or 0. In the previous formula, the calculation is made according to the service number assigned sequentially from 1. When the service number is assigned from 0 , the calculation must be done adding 1 to the service number. Prior to this step, when the service data is sent to the PCIE device, the host computer side sends a service number corresponding to the service data to the PCIE device. A service pack service number is determined when the service is established, and the PCIE device can store the service number for use.

Step 4. The virtual machine obtains, according to the CLAT entry base address and the entry offset, a CLAT entry corresponding to the service to be processed.

Step 5. The virtual machine sends, to the PCIE device, operating information in a block of memory corresponding to the CLAT entry corresponding to the service to be processed.

Furthermore, at this stage, the virtual machine writes, in a cache memory of the PCIE device, the operating information of the memory block, and the PCIE device reads, from the cache memory, the operating information of the memory block, thereby improving the efficiency of reading data and increasing, in addition, the speed of service processing.

In addition, when the BAT and CLAT input is sent to the PCIE device, the virtual machine can write the BAT and CLAT entries into the cache memory of the PCIE device for later use by the PCIE device.

Fourth, the PCIE device receives the operation information sent by the virtual machine, where the operating information is obtained by means of the virtual machine from a block of memory corresponding to the request message.

Way of implementation 2:

In the first place, the PCIE device reads, according to the BAT address corresponding to the service to be processed, the BAT corresponding to the service to be processed.

Secondly, the PCIE device determines, according to the type of service of the service to be processed, a base CLAT entry address corresponding to the type of service of the service to be processed of the BAT corresponding to the service to be processed.

Third, the PCIE device determines an entry displacement according to the service number of the service to be processed.

In this step, the manner of determining the entry offset is the same as in the implementation manner 1, and the details are not described again in this document.

Fourth, the PCIE device reads a CLAT entry corresponding to the entry offset and the CLAT entry base address that corresponds to the service type of the service to be processed.

Fifth, the PCIE device reads operation information in a block of memory corresponding to the CLAT input.

In this step, the PCIE device can read, from a cache memory of the PCIE device, the operating information of the memory block for later use, thereby improving the service processing speed.

In this embodiment of the present invention, the implementation mode 2 can be executed by a read / write module of direct memory access (in English: Direct Memory Access, DMA for short) of the PCIE device.

Step 407. The PCIE device processes the service information in accordance with the operating information. Specifically, a packet is processed by a VF corresponding to the previous function number.

Step 408. A migration management device in a physical home computer migrates directly a memory block, a BAT and a CLAT that are in a physical central computer of origin to a physical host computer during the live migration.

The physical host computer and the physical home computer are connected to different peripheral interconnection devices of peripheral components, PCIE. The PCIE device is a hardware device that supports the single-root I / O virtualization standard, SR-IOV. The PCIE device has at least one physical function, PF, and at least one virtual function, VF, and a unique function number is configured for each PF and each VF separately.

In addition, the migration management device in the physical home computer writes, in a memory of the physical host computer, the data that is in a cache memory of a PCIE device connected to the physical home computer.

Step 409. A migration management device in the physical host computer configures a correspondence between a base address table address, BAT, of a virtual machine of the physical home computer of origin and a function number corresponding to a machine. Migrated virtual in a VCT virtual machine configuration table of a PCIE device connected to the target physical host.

Specifically, the migration management device in the destination physical host obtains the address of the base address table, BAT, of the virtual machine of the physical home computer of origin and the function number (the function number can be specified by the virtual machine or assigned by a VMM) corresponding to the virtual machine of the migrated destination. The migration management device in the destination physical central computer sends the BAT address and the corresponding function number to the migrated destination virtual machine to the PCIE device. The PCIE device connected to the physical host computer stores the correspondence between the address of the bAt and the function number in the virtual machine configuration table, VCT.

In this embodiment, the migration management device can be a VMM.

In this embodiment of the present invention, a virtual machine assigns a memory to at least one service located in a PCIE device, generates a BAT and a CLAT according to the memory assigned to each service, and sends a BAT address. and a function number to the PCIE device, so that the PCIE device can register a correspondence between the BAT address and the function number. During the processing of services, the PCIE device obtains, from the virtual machine, operating information of a service according to the correspondence between the BAT address and the function number. This is easy to do and requires less time. During the live migration of a virtual machine the problems caused by the migration of records are avoided, provided that a block of memory, a BAT and a CLAT are migrated directly to a physical machine of destination, and that a correspondence between a number of function corresponding to a migrated virtual machine and a BAT address in a virtual machine of a physical central computer of this origin configured in a VCT in a PCIE device after the migration. This is simple and easy to implement, greatly increases the speed of live migration of a virtual machine and reduces downtime. Form 5

This embodiment of the present invention provides a physical central computer, which is applied with the method provided in the embodiment 1. Referring to FIG. 8, a virtual machine 510 is executed in the physical central computer 51, and the physical central computer 51 is connected to an peripheral interconnect device, PCIE, 52. The PCIE 52 device is a hardware device that supports the standard Virtualization I / O of a single root, SR-IO ^v . The PCIE device 52 has at least one physical function, PF, 521 and at least one virtual function, VF, 522, and a unique function number is configured for each PF 521 and each VF 522 separately. The virtual machine 510 includes:

an allocation module 511, configured to allocate a memory to at least one service located in the PCIE device, where the memory allocated to each service includes multiple memory blocks, each of the multiple blocks of memory being used to store operating information of a corresponding service, and each memory block is a section of memory space in which the physical guest addresses are consecutive;

an input generation module 512, configured to generate a table of base addresses, BAT, and a table of logical chip addresses, CLAT, according to the memory assigned to each service, where the BAT includes a base address of CLAT corresponding to each service, and the CLAT includes a first address of each memory block; Y

a send module 513, configured to send to the PCIE device a BAT address and a function number corresponding to the virtual machine, where the PCIE device is configured to: register a correspondence between the BAT address and the function number , and obtain, from the virtual machine, operating information of a service according to the correspondence between the BAT address and the function number.

In this embodiment of the present invention, a virtual machine assigns a memory to at least one service located in a PCIE device, generates a BAT and a CLAT according to the memory assigned to each service, and sends a BAT address. and a function number to the PCIE device, so that the PCIE device can register a correspondence between the BAT address and the function number. During the processing of services, the PCIE device obtains, from the virtual machine, operating information of a service according to the correspondence between the BAT address and the function number. This is easy to do and requires less time. During the live migration of a virtual machine the problems caused by the migration of records, whenever a block of memory, a BAT and a CLAT are migrated directly to a physical machine of destination, and that a correspondence between a number of function corresponding to a virtual machine migrated and a BAT address in a virtual machine of a physical central computer of origin this configured in a VCT in a PCIE device after the migration. This is simple and easy to implement, greatly increases the speed of live migration of a virtual machine and reduces downtime. Form of realization 6

This embodiment of the present invention provides a physical central computer, which is applied with the method provided in the embodiment 4. Referring to FIG. 9, a virtual machine 610 is executed in the physical central computer 61, and the physical central computer 61 is connected to an peripheral interconnection device, PCIE, 62. The PCIE 62 device is a hardware device that supports the standard Virtualization I / O of a single root, SR-IOv. The PCIE device 62 has at least one physical function, PF, 621 and at least one virtual function, VF 622, and a unique function number is configured for each PF 621 and each VF 622 separately. The virtual machine 610 includes an assignment module 611, an input generation module 612 and a delivery module 613.

The assignment module 611 is configured to allocate a memory to at least one service located in the PCIE device, where the memory allocated to each service includes multiple blocks of memory, each of the multiple blocks of memory being used to store operating information. of a corresponding service, and each memory block is a section of memory space in which the physical guest addresses are consecutive.

When the virtual machine 610 is started, the virtual machine 610 is applied to a memory of a physical central computer for the PCIE 62 device and allocates the memory that is applied to a service. Specifically, after receiving an application from the virtual machine, a PCIE device driver of the physical central computer 61 allocates a memory to the virtual machine. The PCIE device driver of the physical central computer 61 refers to a driver that is used to operate the PCIE device and that is installed on a hard disk of the physical central computer.

In this embodiment, the operation information of the service can be configuration information, a context table, a service resource or a table about the environment. In the context table, various states of the service are recorded, such as an execution status, a queuing state, a timer status and a statistics collection state.

In this embodiment, the memory blocks assigned by the virtual machine 610 to each service are of a constant size and include consecutive GPAs, that is, the sizes of the multiple memory blocks assigned to each service are identical. Specifically, a service can occupy independently a block of memory, or multiple services can occupy a block of memory together. In addition, in this embodiment, an unimportant size of the memory block is a size of a memory page and is generally 4K.

The input generation module 612 is configured to generate a table of base addresses, BAT, and a table of logical chip addresses, CLAT, according to the memory assigned to each service, where the BAT includes a base address of CLAT corresponding to each service, and the CLAT includes a first address of each memory block. Specifically, a first address of the BAT may be a physical address of the physical central computer (HPA), or a physical address of the virtual machine (GPA). If the first address of the BAT is the GPA, when the address is accessed, a PCIE adapter needs to implement a conversion from GPA to HPA through the use of VT-D (an Intel technology) or IOMMU (an AMD technology). on a CPU, or other technology that can automatically implement the conversion of addresses from GPA to HPA.

If there are several services of the same type of service in the CLAT, to differentiate between the different services of the same type of service, a service number can be assigned to each service. The service number can be used for the calculation of a CLAT input offset, and the details are described below. In the present document, the type of service is a type of a service that can be executed by the PCIE 62 device, for example, a divert service, a segmentation service and a checksum service, which is executed by a network adapter.

In addition, the assignment module 611 can allocate a block of memory to the BAT and the CLAT. A number of bits in a first address of the memory block is constant, and therefore, a number of octets in the first address of the memory block is also constant. For example, the number of bits in the first address of the memory block is 64, and a number of octets occupied by a first address of each memory block is 8B. Therefore, if a size of a memory assigned to the CLA ^t is 4K, a maximum number of entries in each CLAT is 4K / 8 = 512. When the entries in a CLAT are not sufficient, it is necessary to generate a CLAT of multiple levels to store a first address of a block of memory.

It is, the CLAT can be a CLAT of a single level or a CLAT of multiple levels. The CLAT of multiple levels includes N levels of CLAT, where N> 2 and N is an integer number. Each level of CLAT includes multiple entries. Each entry in the nth level of CLAT is used to indicate a direction of the level (n + 1) -third of CLAT, where 2 <n <n + 1 <N and n is an integer number. Each entry in the N-th level of CLAT is used to indicate a first address of one of the memory blocks.

As shown in FIG. 6, in the CLAT of a single level, each entry of the CLAT corresponds to a block of memory.

As shown in FIG. 7, in the CLAT of multiple levels, using a CLAT of two levels as an example, an entry in the first level of CLAT corresponds to multiple entries in the second level of CLAT, and an entry in the second level of CLAT corresponds to a block of memory.

Therefore, the BAT may also include a size of a CLAT entry (a size of a memory block allocated to the CLAT), the size of the memory block, a number of levels of the CLAT and the like.

The sending module 613 is configured to send to the PCIE device a BAT address and a function number corresponding to the virtual machine, where the PCIE device is configured to: register a correspondence between the BAT address and the function number. , and obtain, from the virtual machine, operating information of a service according to the correspondence between the BAT address and the function number.

The address of the BAT can be a first address of the BAT. The function number is a function number of a PF or VF corresponding to the virtual machine. The function number of the PF or the VF corresponding to the virtual machine can be specified by the virtual machine or by a VMM.

In addition, the virtual machine 610 may include a reception module 614.

The reception module 614 is configured to receive a request message that is used to obtain operation information of a service to be processed and that is sent by the PCIE device, where the request message includes the address of the BAT, a type of Service of the service to be processed and a service number of the service to be processed.

The sending module 613 is further configured to send, to the PCIE device according to the request message, operation information in a memory block corresponding to the service to be processed.

In this embodiment, the sending module 613 can be specifically configured to:

obtain the BAT corresponding to the address of the BAT; determine, according to the type of service of the service to be processed and the BAT, a base entry address of CLAT corresponding to the type of service; determine an entry displacement according to the service number of the service to be processed; obtain, in accordance with the CLAT entry base address and the entry offset, a CLAT entry corresponding to the service to be processed; and sending, to the PCIE device, operating information in a block of memory corresponding to the CLAT entry corresponding to the service to be processed.

In addition, the virtual machine 610 writes, in a cache memory of the PCIE device, the operating information in the memory block, and the PCIE device reads, from the cache memory, the operating information in the memory block, thereby improving the efficiency of reading data and increasing, in addition, the speed of service processing.

Further, when the BAT and CLAT input is sent to the PCIE device, the virtual machine 610 can write the BAT and CLAT input into the cache memory of the PCIE device for later use by the PCIE device.

In addition, the sizes of multiple blocks of memory assigned to the service to be processed are identical.

The shipping module can be configured specifically to:

when the CLAT is a CLAT of a single level, multiply the service number of the service to be processed by a size of a memory occupied by the operating information of the service to be processed, and divide the result of the multiplication by the size of the block of memory assigned to the service to process, to obtain a quotient and a remainder for the CLAT of a single level, where when the rest for the CLAT of a single level is greater than 0, the entry displacement is equal to the quotient for the CLAT of a single level, or when the rest for the CLAT of a single level is equal to 0, the entry displacement is equal to the quotient for the CLAT of a single level minus 1; or

when the CLAT is a CLAT of multiple levels, where the entry displacement includes an entry displacement of each level of CLAT, calculate an entry displacement of the X-th level of CLAT in the following way:

where when a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the x-th level of CLAT is equal to 0, the entry displacement of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the operation information of the service to be processed, B is the size of the memory block assigned to the service to be processed, and C is the number of entries in the N-th level of CLAT.

When the input displacement of the CLAT x-th level is 0, it indicates that a corresponding entry is the first entry in the x-th level of CLAT, when the entry offset of the x-th level of CLAT is 1, it indicates that a corresponding entry is the second entry in the x-th level of CLAT, and so on. It should be noted that a remainder for the N-th level of CLAT (ie, the last CLAT level) is used to indicate the operation information, corresponding to the service number, in a given memory block. When the remainder for the Nth level of CLAT is equal to 0, the operation information corresponding to the service number is the last element of operation information in the given memory block. When the remainder for the N-th level of CLAT is greater than 0, the operation information corresponding to the service number is the m-th element of operating information in the given memory block, and m is equal to the rest for the level N-th of CLAT.

As for a specific example, reference is made to step 406 in the embodiment 4. Detailed descriptions are omitted in the present document.

In this embodiment of the present invention, a virtual machine assigns a memory to at least one service located in a PCIE device, generates a BAT and a CLAT according to the memory assigned to each service, and sends a BAT address. and a function number to the PCIE device, so that the PCIE device can register a correspondence between the BAT address and the function number. During the processing of services, the PCIE device obtains, from the virtual machine, operating information of a service according to the correspondence between the BAT address and the function number. This is easy to do and requires less time. During the live migration of a virtual machine the problems caused by the migration of records are avoided, provided that a block of memory, a BAT and a CLAT are migrated directly to a physical machine of destination, and that a correspondence between a number of function corresponding to a migrated virtual machine and a BAT address in a virtual machine of a physical central computer of this origin configured in a VCT in a PCIE device after the migration. This is simple and easy to implement, greatly increases the speed of live migration of a virtual machine and reduces downtime. Form of realization 7

This embodiment of the present invention provides a PCIE adapter, which is applied with the method provided in embodiment 2. Referring to FIG. 10, the PCIE device 71 is connected to a physical central computer 72, and at least one virtual machine 720 is executed in the physical central computer 72. The PCIE device is a hardware device that satisfies the I / O virtualization standard of a single root, SR-IOV. The PCIE device 71 has at least one physical function PF 711 and at least one virtual function VF 712, and a unique function number is configured for each PF 711 and each VF 712 separately. Each virtual machine 720 includes a table of base addresses, BAT, and the BAT is generated by a virtual machine, to which the BAT belongs, according to a memory assigned to at least one service located in the PCIE device. The memory allocated to each service includes multiple blocks of memory, each of the multiple blocks of memory is used to store operating information of a corresponding service, and each memory block is a section of memory space in which the physical addresses of guest are consecutive. The BAT includes a base CLAT entry address corresponding to each service, and a CLAT includes a first address of each memory block. The PCIE 71 device also includes:

a reception module 713, configured to receive a address from the base address table, BAT, in each virtual machine and a function number corresponding to each virtual machine; Y

a generation module 714, configured to store a correspondence between the address of the BAT and the function number in a virtual machine configuration table, VCT, so that the VCT includes the correspondence between the function number and the address of the the BAT.

In this embodiment of the present invention, a PCIE device receives a BAT address and a function number, and stores a correspondence between the address of the BAT and the function number in a virtual machine configuration table, v Ct. During the processing of services, the PCIE device obtains, from a virtual machine, operating information of a service according to the correspondence between the BAT address and the function number. This is easy to do and requires less time. During the live migration of a virtual machine the problems caused by the migration of records are avoided, provided that a block of memory, a BAT and a CLAT are migrated directly to a physical host computer of destination, and that a correspondence between a number of function corresponding to a virtual machine migrated and a BAT address in a virtual machine of a physical central computer of origin this configured in a VCT in a PCIE device after the migration. This is simple and easy to implement, greatly increases the speed of live migration of a virtual machine and reduces downtime.

Form of realization 8

This embodiment of the present invention provides a PCIE adapter, which is applied with the method provided in the embodiment 4. With reference to FIG. 11, the PCIE device 81 is connected to a physical central computer 82, and at least one virtual machine 820 is executed in the physical central computer 82. The PCIE device is a hardware device that satisfies the I / O virtualization standard of a single rafz, SR-IOV. The PCIE device 81 has at least one physical function, PF 811 and at least one virtual function, VF 812, and a unique function number is configured for each PF 811 and each VF 812 separately. Each virtual machine 820 includes a table of base addresses, BAT, and the BAT is generated by a virtual machine, to which the BAT belongs, according to a memory allocated to at least one service located in the PCIE device. The memory allocated to each service includes multiple blocks of memory, each of the multiple blocks of memory is used to store operating information of a corresponding service, and each memory block is a section of memory space in which physical addresses of guest are consecutive. The BAT includes a CLAT entry base address corresponding to each service, and the CLAT includes a first address of each memory block. The PCIE 81 device also includes:

a reception module 813, configured to receive a address from the base address table, BAT, in each virtual machine and a function number corresponding to each virtual machine; Y

a generation module 814, configured to store a correspondence between the address of the BAT and the function number in a virtual machine configuration table, VCT, so that the VCT includes the correspondence between the function number and the address of the the BAT.

The VCT is stored in the PCIE device (for example, a network adapter), and the BAT and CLAT are saved in the virtual machine. Specifically, the VCT can be generated in advance, and after receiving the BAT address and the function number, the PCIE device writes the address of the BAT and the function number in the VCT. More specifically, the VCT can be generated by a VMM or the PCIE.

Specifically, the address of the BAT can be an HPA or a GPA. If the address of the BAT is the GPA, when the address is accessed, the PCIE device needs to implement a conversion from GPA to HPA through the use of VT-D (an Intel technology) or IOMMU (an AMD technology) in a CPU, or other technology that can automatically implement the conversion of addresses from GPA to HPA.

In addition, the PCIE device 81 may further include a determination module 815 and a obtaining module 816.

The determination module 815 is configured to determine, in accordance with service information of a service to be processed and the VCT, a BAT address corresponding to the service to be processed.

The service information includes service information sent by a central computer side or service information sent by a network side. The service information sent by the central computer side is generally service data. The service information sent by the network side is generally a service pack.

The obtaining module 816 is configured to obtain, from a memory corresponding to the service to be processed, operating information of the service to be processed according to the BAT address corresponding to the service to be processed, a service type of the service to be processed and a service number of the service to be processed.

In this embodiment of the present invention, the reception module 813 can also be configured to receive the service information sent by the central computer side or the network side.

The determination module 815 can be specifically configured to:

determine a function number corresponding to the service to be processed, in accordance with a feature of the service information, sent by the central computer side, of the service to be processed or in accordance with a PCIE interface used when the service information is received, transmitted by the central computer side of the service to be processed, where the feature is used to indicate the function number corresponding to the service to be processed; Y

determine, in accordance with the VCT and the function number determined corresponding to the service to be processed, the address of the BAT corresponding to the function number corresponding to the service to be processed.

In this embodiment, a feature in the service packet sent by the network side may be a field in the service pack. A field used as a feature in this document may vary with a type of a PCIE device. For example, when the PCIE device is a network adapter, a destination MAC address can be used as a feature. That is, when a service packet is received, the network adapter can determine a corresponding function number according to the destination MAC address in the service pack. In a case in which the central computer side sends the service data, a corresponding function number can be determined according to a PCIE interface used when the service data transmitted by the central computer side is received.

The central computer side refers to a system that includes a CPU, a memory and a hard disk that are from a physical central computer. The network side refers to an external network side connected to the physical central computer. In a manner of implementing this embodiment of the present invention, the obtaining module 816 can be specifically configured to:

send a request message to a virtual machine corresponding to the function number determined that corresponds to the service to be processed, where the request message includes the address of the BAT corresponding to the function number corresponding to the service to be processed, the type of service of the service to be processed and the service number of the service to be processed; and receive operating information of the service to be processed, sent by the virtual machine corresponding to the function number corresponding to the service to be processed, where the operation information of the service to be processed is obtained from a memory block corresponding to the request message through the virtual machine corresponding to the function number that corresponds to the service to be processed.

The type of service is a type of a service corresponding to the service pack or service data received by the PCIE device. The service corresponding to the service package is a service transported in the service package. The service corresponding to the service data is a service transported in a service package that is obtained after the PCIE device processes the service data.

In another embodiment of this embodiment of the present invention, the obtaining module 816 can be specifically configured to:

read, according to the address of the BAT corresponding to the service to be processed, the BAT corresponding to the service to be processed; determine, according to the type of service of the service to be processed, a base CLAT entry address corresponding to the type of service of the service to be processed of the BAT corresponding to the service to be processed; determine an entry displacement according to the service number of the service to be processed; read a CLAT entry corresponding to the entry displacement and the CLAT entry base address that corresponds to the service type of the service to be processed; and read operation information in a block of memory corresponding to the CLAT input.

In this embodiment of the present invention, the CLAT is a single level CLAT or a multiple level CLAT. The CLAT of multiple levels includes N levels of CLAT, where N> 2 and N is an integer number. Each level of CLAT includes multiple entries. Each entry in the nth level of CLAT is used to indicate a direction of the level (n + 1) -third of CLAT, where 2 <n <n + 1 <N and n is an integer number. Each entry in the N-th level of CLAT is used to indicate a first address of one of the memory blocks.

In this embodiment of the present invention, the sizes of the multiple memory blocks assigned to the service to be processed are identical.

The obtaining module 816 can be configured specifically for:

when the CLAT is a CLAT of a single level, multiply the service number of the service to be processed by a size of a memory occupied by the operating information of the service to be processed, and divide the result of the multiplication by the size of the block of memory assigned to the service to process, for obtain a quotient and a remainder for the CLAT of a single level, where when the rest for the CLAT of a single level is greater than 0, the entry displacement is equal to the quotient for the CLAT of a single level, or when the rest for the CLAT of a single level is equal to 0, the entry displacement is equal to the quotient for the CLAT of a single level minus 1; or

when the CLAT is a CLAT of multiple levels, where the entry displacement includes an entry displacement of each level of CLAT, calculate an entry displacement of the X-th level of CLAT in the following way:

where when a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the x-th level of CLAT is equal to 0, the entry displacement of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the operation information of the service to be processed, B is the size of the memory block assigned to the service to be processed, and C is the number of entries in the N-th level of CLAT.

When the input displacement of the CLAT x-th level is 0, it indicates that a corresponding entry is the first entry in the x-th level of CLAT, when the entry offset of the x-th level of CLAT is 1, it indicates that a corresponding entry is the second entry in the x-th level of CLAT, and so on. It should be noted that a remainder for the N-th level of CLAT (ie, the last CLAT level) is used to indicate the operation information, corresponding to the service number, in a given memory block. When the remainder for the Nth level of CLAT is equal to 0, the operation information corresponding to the service number is the last element of operation information in the given memory block. When the remainder for the N-th level of CLAT is greater than 0, the operation information corresponding to the service number is the m-th element of operating information in the given memory block, and m is equal to the rest for the level N-th of CLAT.

The calculation of the input offset is described below by means of an example. For example, a size of each used memory is 1 KB, a size of a memory block is 4 KB, a number of bits in a first address of a memory block is 64, and a size of a block of memory is 64. memory assigned to a CLAT is also 4 KB.

As for a specific example, reference is made to step 406 in the embodiment 4. Detailed descriptions are omitted in the present document.

In this embodiment of the present invention, a PCIE device receives a BAT address and a function number, and stores a correspondence between the address of the BAT and the function number in a virtual machine configuration table, v Ct. During the processing of services, the PCIE device obtains, from a virtual machine, operating information of a service according to the correspondence between the BAT address and the function number. This is easy to do and requires less time. During the live migration of a virtual machine the problems caused by the migration of records are avoided, provided that a block of memory, a BAT and a CLAT are migrated directly to a physical host computer of destination, and that a correspondence between a number of function corresponding to a virtual machine migrated and a BAT address in a virtual machine of a physical central computer of origin this configured in a VCT in a PCIE device after the migration. This is simple and easy to implement, greatly increases the speed of live migration of a virtual machine and reduces downtime.

Form of realization 9

This embodiment of the present invention provides a migration management device. The migration management device is arranged in a physical host computer and is applied with the procedure provided in the embodiment 3. Referring to FIG. 12, the migration management device includes:

a first obtaining module 901, configured to obtain a address from a base address table, BAT, in a virtual machine of a physical home computer during a live migration, where the physical host computer and the physical central computer of origin are connected to Different peripheral interconnection devices of peripheral components, PCIE, the PCIE device is a hardware device that supports the single-root I / O virtualization standard, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, where a unique function number is configured for each PF and each VF separately;

a second obtaining module 902, configured to obtain a function number corresponding to a migrated virtual machine; Y

a configuration module 903, configured to configure a correspondence between the address of the BAT and the function number in a virtual machine configuration table, VCT, in a PCIE device connected to the physical host computer.

In this embodiment of the present invention, during the live migration of a virtual machine the problems caused by the migration of records are avoided, provided that a block of memory, a BAT and a CLAT are automatically and directly migrated to a physical machine of destination during a migration process, and that a correspondence between a number of function corresponding to a virtual machine migrated and a BAT address in a virtual machine of a physical central computer of origin this configured in a VCT in a PCIE device after migration. This is simple and easy to implement, greatly increases the speed of live migration of a virtual machine and reduces downtime.

It should be noted that when the virtual machine provided in the previous embodiments manages a memory of the virtual machine, the division of the above functional modules is only an example for the description. In a real application, the above functions can be implemented by different functional modules according to a requirement, or a function thereof can be implemented by more modules, or multiple functions thereof are implemented by a module. That is, an internal structure of a device is divided into different functional modules to implement all or some of the functions described above. Furthermore, the physical central computer provided in the previous embodiments and the embodiments of the procedure for managing a memory of a virtual machine belong to the same idea. As for a specific implementation process thereof, reference is made to the procedural embodiments, and the details are not described again in this document.

The sequence numbers of the previous embodiments of the present invention are for illustrative purposes only and are not intended to indicate priorities of the embodiments.

One skilled in the art can understand that all or some of the stages of the embodiments can be implemented by hardware or a program that gives orders to a related hardware. The program can be stored in a computer readable storage medium. The storage medium may include: a read-only memory, a magnetic disk, an optical disk or the like.

The above descriptions are merely examples of embodiments of the present invention and are not intended to limit the present invention.

Claims (24)

1. A procedure for managing a memory of a virtual machine, in which the virtual machine is executed in a physical central computer, the physical central computer is connected to an peripheral interconnection device of peripheral components, PCIE, the PCIE device is a hardware device that supports the single-root I / O virtualization standard, SR-lOv, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, in which a function number only one is configured for each PF and each VF separately; wherein the virtual machine executes a service using a VF of the at least one VF, and the virtual machine corresponds to a function number of the VF of the at least one VF; and the procedure comprises: assign, by means of the virtual machine, a memory to at least one service located in the PCIE device, where the memory assigned to each service comprises multiple blocks of memory, each one of the multiple memory blocks is used to store operating information of a corresponding service, and each memory block is a section of memory space in which physical guest addresses are consecutive; generate, through the virtual machine, a table of base addresses, BAT, and a table of logical chip addresses, CLAT, according to the memory assigned to each service, where the BAT comprises a base entry address of CLAT corresponding to each service, and the CLAT comprises a first address of each memory block; send to the PCIE device, through the virtual machine, a BAT address and a function number corresponding to the virtual machine, where a correspondence between the BAT address and the function number corresponding to the virtual machine is stored in a table of configuration of virtual machine, VCT, of the PCIE device connected to the physical central computer; wherein when the virtual machine is migrated from the physical central computer to a physical host computer, the memory blocks, the BAT and the CLAT are migrated to the virtual machine migrated in the physical host computer of destination; Y in which after the migration, a correspondence between the address of the BAT and a number of function corresponding to the migrated virtual machine, and is of a VF in a PCIE device connected to the physical central computer of destination, is configured in a v Ct on the PCIE device connected to the target physical host. The method according to claim 1, wherein the at least one service comprises a service to be processed, and the method further comprises: receiving a request message that is used to obtain operating information of the service to be processed and sent by the PCIE device, where the request message comprises the address of the BAT, a service type of the service to be processed and a number of service of the service to be processed; and sending, to the PCIE device according to the request message, operation information in a block of memory corresponding to the service to be processed. 3. The method according to claim 2, wherein sending, to the PCIE device according to the request message, operation information in a memory block corresponding to the service to be processed comprises: obtain the BAT corresponding to the address of the BAT; determine, according to the type of service of the service to be processed and the BAT, a base entry address of CLAT corresponding to the type of service; determine an entry displacement according to the service number of the service to be processed; obtain, in accordance with the CLAT entry base address and the entry offset, a CLAT entry corresponding to the service to be processed; Y send, to the PCIE device, operating information in a block of memory corresponding to the CLAT input corresponding to the service to be processed. 4. The method according to claim 3, wherein the CLAT is a CLAT of a single level or a CLAT of multiple levels; the CLAT of multiple levels comprises N levels of CLAT, where N> 2 and N is an integer number; each level of CLAT comprises multiple entries; each entry in the nth level of CLAT is used to indicate a direction of the level (n + 1) -this CLAT, where 2 <n <n + 1 <N, and n is an integer; and each entry in the N-th level of CLAT is used to indicate a first address of one of the memory blocks. The method according to claim 4, wherein the sizes of the multiple memory blocks assigned to the service to be processed are identical; Y The determination of an entry displacement according to the service number of the service to be processed comprises: When the CLAT is a CLAT of a single level, multiply the service number of the service to be processed by a size of a memory occupied by the operation information of the service to be processed, and divide the result of the multiplication by the size of the memory block assigned to the service to be processed, to obtain a quotient and a remainder for the CLAT of a single level, where when the rest for the c The T of a single level is greater than 0, the offset of input is equal to the quotient for the CLAT of a single level, or when the rest for the CLAT of a single level is equal to 0, the entry displacement is equal to the quotient for the CLAT of a single level minus 1; or when the CLAT is a CLAT of multiple levels, where the entry displacement comprises an entry displacement of each level of CLAT, calculate an entry displacement of the X-th level of CLAT in the following way:

where when a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the x-th level of CLAT is equal to 0, the entry displacement of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the operation information of the service to be processed, B is the size of the memory block assigned to the service to be processed and C is the number of entries in the N-th level of CLAT. 6. A method for configuring an peripheral interconnect device of peripheral components, PCIE, where the PCIE device is connected to a physical central computer, at least one virtual machine is running on the physical central computer, the PCIE device is a hardware device which supports the single-root I / O virtualization standard, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, where a unique function number is configured for each PF and each VF separately; each virtual machine comprises a table of base addresses, BAT, where the BAT is generated by a virtual machine, to which the BAT belongs, according to a memory assigned to at least one service located in the PCIE device; the memory allocated to each service comprises multiple blocks of memory, each of the multiple blocks of memory is used to store operating information of a corresponding service, each memory block is a section of memory space in which physical addresses of guest they are consecutive; the BAT comprises a base entry address of chip logic address table, CLAT, corresponding to each service, and the CLAT comprises a first address of each memory block; Y The procedure includes: receive, through the PCIE device, a address from the base address table, BAT, in each virtual machine and a function number corresponding to each virtual machine; save a correspondence between the address of the BAT and the function number in a virtual machine configuration table, VCT; Y wherein a virtual machine executes at least one service using a VF of the at least one VF, and the virtual machine corresponds to a function number of the VF of the at least one VF; when the virtual machine is migrated from the physical central computer to a physical host computer, the memory blocks, the BAT and the CLAT of the virtual machine are migrated to the virtual machine migrated in the physical host computer of destination; and after the migration, a correspondence between the address of the BAT of the virtual machine and a number of function corresponding to the migrated virtual machine, and is of a VF in a PCIE device connected to the physical central computer of destination, is configured in a VCT on the PCIE device connected to the target physical host. The method according to claim 6, wherein the at least one service comprises a service to be processed, and the method further comprises: determine, in accordance with the service information of the service to be processed and the VCT, a BAT address corresponding to the service to be processed; Y obtain, from a memory corresponding to the service to be processed, operating information of the service to be processed in accordance with the BAT address corresponding to the service to be processed, a service type of the service to be processed and a service number of the service to be processed. process. The method according to claim 7, wherein determining, in accordance with service information of the service to be processed and the VCT, a BAT address corresponding to the service to be processed comprises: receive the service information sent by one side of the central computer or one network side of the service to process; determine a function number corresponding to the service to be processed, in accordance with a feature of the service information, sent by the central computer side, of the service to be processed or in accordance with a PCIE interface used when the service information is received, transmitted by the central computer side of the service to be processed, where the feature is used to indicate the function number corresponding to the service to be processed; Y determine, in accordance with the VCT and the function number determined corresponding to the service to be processed, the address of the BAT corresponding to the function number corresponding to the service to be processed. 9. The method according to claim 8, wherein obtaining, from a memory corresponding to the service to be processed, operation information of the service to be processed in accordance with the BAT address corresponding to the service to be processed, a type of Service of the service to be processed and a service number of the service to be processed comprises: send a request message to a virtual machine corresponding to the determined function number corresponding to the service to be processed, where the request message comprises the address of the BAT corresponding to the function number corresponding to the service to be processed, the type of service of the service to be processed and the service number of the service to be processed; Y receive operating information of the service to be processed, sent by the virtual machine corresponding to the function number corresponding to the service to be processed, where the operating information of the service to be processed is obtained from a block of memory corresponding to the message of request by means of the virtual machine corresponding to the number of function that corresponds with the service to be processed. The method according to claim 7 or 8, wherein obtaining, from a memory corresponding to the service to be processed, operation information of the service to be processed in accordance with the BAT address corresponding to the service to be processed, a type of service of the service to be processed and a service number of the service to be processed comprises: read, according to the address of the BAT corresponding to the service to be processed, the BAT corresponding to the service to be processed; determine, according to the type of service of the service to be processed, a base CLAT entry address corresponding to the type of service of the service to be processed of the BAT corresponding to the service to be processed; determine an entry displacement according to the service number of the service to be processed; read a CLAT entry corresponding to the entry displacement and the CLAT entry base address that corresponds to the service type of the service to be processed; Y read operation information in a block of memory corresponding to the CLAT input. The method according to claim 10, wherein the CLAT is a single level CLAT or a multiple level CLAT; the CLAT of multiple levels comprises N levels of CLAT, where N> 2 and N is an integer number; each level of CLAT comprises multiple entries; each entry in the nth level of CLAT is used to indicate a direction of the level (n + 1) -this CLAT, where 2 <n <n + 1 <N, and n is an integer; and each entry in the N-th level of CLAT is used to indicate a first address of one of the memory blocks. The method according to claim 11, wherein the sizes of the multiple memory blocks assigned to the service to be processed are identical; Y The determination of an entry displacement according to the service number of the service to be processed comprises: when the CLAT is a CLAT of a single level, multiply the service number of the service to be processed by a size of a memory occupied by the operating information of the service to be processed, and divide the result of the multiplication by the size of the block of memory assigned to the service to process, to obtain a quotient and a remainder for the CLAT of a single level, where when the rest for the CLAT of a single level is greater than 0, the entry displacement is equal to the quotient for the CLAT of a single level, or when the rest for the CLAT of a single level is equal to 0, the entry displacement is equal to the quotient for the CLAT of a single level minus 1; or when the CLAT is a CLAT of multiple levels, where the entry displacement comprises an entry displacement of each level of CLAT, calculate an entry displacement of the X-th level of CLAT in the following way:

where when a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the x-th level of CLAT is equal to 0, the entry displacement of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the operation information of the service to be processed, B is the size of the memory block assigned to the service to be processed and C is the number of entries in the N-th level of CLAT. 13. A physical central computer, in which a virtual machine is executed in the physical central computer, the physical central computer is connected to an express interconnection device of peripheral components, PCIE, the PCIE device is a hardware device that supports the single-root I / O virtualization standard, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, in which a unique function number is configured for each PF and each VF separately; wherein the virtual machine executes a service using a VF of the at least one VF, and the virtual machine corresponds to a function number of the VF of the at least one VF; and the virtual machine comprises: an allocation module, configured to allocate a memory to at least one service located in the PCIE device, where the memory allocated to each service comprises multiple blocks of memory, each of the multiple blocks of memory being used to store operating information of a corresponding service, and each memory block is a section of memory space in which physical guest addresses are consecutive; a module of generation of inputs, configured to generate a table of base addresses, BAT, and a table of logical addresses of chip, CLAT, according to the memory assigned to each service, where the BAT comprises a base entry address of CLAT corresponding to each service, and the CLAT comprises a first address of each memory block; Y a sending module, configured to send to the PCIE device a BAT address and a function number corresponding to the virtual machine, in which a correspondence between the address of the BAT and the function number corresponding to the virtual machine is saved in a virtual machine configuration table, VCT, of the PCIE device connected to the physical central computer; wherein when the virtual machine is migrated from the physical central computer to a physical host computer, the memory blocks, the BAT and the CLAT are migrated to the virtual machine migrated in the physical host computer of destination; Y in which after the migration, a correspondence between the address of the BAT and a number of function corresponding to the migrated virtual machine, and is of a VF in a PCIE device connected to the physical central computer of destination, is configured in a v Ct on the PCIE device connected to the target physical host. 14. The physical central computer according to claim 13, wherein the at least one service comprises a service to be processed, and the virtual machine further comprises: a reception module, configured to receive a request message that is used to obtain operation information of the service to be processed and which is sent by the PCIE device, where the request message comprises the address of the BAT, a service type of the service to be processed and a service number of the service to be processed; Y the sending module is also configured to send, to the PCIE device according to the request message, operation information in a memory block corresponding to the service to be processed. 15. The physical central computer according to claim 14, wherein the sending module is specifically configured to: obtain the BAT corresponding to the address of the BAT; determine, according to the type of service of the service to be processed and the BAT, a base entry address of CLAT corresponding to the type of service; determine an entry displacement according to the service number of the service to be processed; obtain, in accordance with the CLAT entry base address and the entry offset, a CLAT entry corresponding to the service to be processed; and sending, to the PCIE device, operating information in a block of memory corresponding to the CLAT entry corresponding to the service to be processed. 16. The physical central computer according to claim 15, wherein the CLAT is a single level CLAT or a multiple level CLAT; the CLAT of multiple levels comprises N levels of CLAT, where N> 2 and N is an integer number; each level of CLAT comprises multiple entries; each entry in the nth level of CLAT is used to indicate a direction of the level (n + 1) -this CLAT, where 2 <n <n + 1 <N, and n is an integer; and each entry in the N-th level of CLAT is used to indicate a first address of one of the memory blocks. 17. The physical central computer according to claim 16, wherein the sizes of the multiple blocks of memory assigned to the service to be processed are identical; Y the shipping module is specifically configured to: when the CLAT is a CLAT of a single level, multiply the service number of the service to be processed by a size of a memory occupied by the operating information of the service to be processed, and divide the result of the multiplication by the size of the block of memory assigned to the service to process, to obtain a quotient and a remainder for the CLAT of a single level, where when the rest for the CLAT of a single level is greater than 0, the entry displacement is equal to the quotient for the CLAT of a single level, or when the rest for the CLAT of a single level is equal to 0, the entry displacement is equal to the quotient for the CLAT of a single level minus 1; or when the CLAT is a CLAT of multiple levels, where the entry displacement comprises an entry displacement of each level of CLAT, calculate an entry displacement of the X-th level of CLAT in the following way:

where when a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the x-th level of CLAT is equal to 0, the entry displacement of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the operation information of the service to be processed, B is the size of the memory block assigned to the service to be processed and C is the number of entries in the N-th level of CLAT. 18. An express interconnection device of peripheral components, PCIE, where the PCIE device is connected to a physical central computer, at least one virtual machine runs on the physical central computer, the PCIE device is a hardware device that satisfies the standard Virtualization I / O of a single root, SR-IOV, and the PCIE device has at least one physical function, PF, and at least one virtual function, VF, where a unique function number is configured for each PF and each VF separately; each virtual machine comprises a table of base addresses, BAT, where the BAT is generated by a virtual machine, to which the BAT belongs, according to a memory assigned to at least one service located in the PCIE device; the memory allocated to each service comprises multiple blocks of memory, each of the multiple blocks of memory is used to store operating information of a corresponding service, each memory block is a section of memory space in which physical addresses of guest they are consecutive; the BAT comprises a base entry address of chip logical address table, CLAT, corresponding to each service, and a CLAT comprises a first address of each memory block; and the PCIE device also comprises: a reception module, configured to receive a address from the base address table, BAT, in each virtual machine and a function number corresponding to each virtual machine; a generation module, configured to store a correspondence between the BAT address and the function number in a virtual machine configuration table, VCT; Y wherein a virtual machine executes at least one service using a VF of the at least one VF, and the virtual machine corresponds to a function number of the VF of the at least one VF; when the virtual machine is migrated from the physical central computer to a physical host computer, the memory blocks, the BAT and the CLAT of the virtual machine are migrated to the virtual machine migrated in the physical host computer of destination; and after the migration, a correspondence between the address of the BAT of the virtual machine and a number of function corresponding to the migrated virtual machine, and is of a VF in a PCIE device connected to the physical central computer of destination, is configured in a VCT on the PCIE device connected to the target physical host. 19. The PCIE device according to claim 18, wherein the at least one service comprises a service to be processed, and the PCIE device further comprises: a determination module, configured to determine, in accordance with the service information of the service to be processed and the VCT, a address of a BAT corresponding to the service to be processed; Y a obtaining module, configured to obtain, from a memory corresponding to the service to be processed, operation information of the service to be processed according to the BAT address corresponding to the service to be processed, a service type of the service to be processed and a service number of the service to be processed. 20. The PCIE device according to claim 19, wherein: the reception module is further configured to receive the service information sent by one side of the central computer or a network side of the service to be processed; Y the determination module is specifically configured to: determine a function number corresponding to the service to be processed, in accordance with a feature of the service information, sent by the central computer side, of the service to be processed or in accordance with a PCIE interface used when the service information is received, transmitted by the central computer side of the service to be processed, where the feature is used to indicate the function number corresponding to the service to be processed; Y determine, in accordance with the VCT and the function number determined corresponding to the service to be processed, the address of the BAT corresponding to the function number corresponding to the service to be processed. 21. The PCIE device according to claim 20, wherein the obtaining module is specifically configured for: send a request message to a virtual machine corresponding to the determined function number corresponding to the service to be processed, where the request message comprises the address of the BAT corresponding to the function number corresponding to the service to be processed, the type of service of the service to be processed and the service number of the service to be processed; and receive operating information of the service to be processed, sent by the virtual machine corresponding to the function number corresponding to the service to be processed, where the operation information of the service to be processed is obtained from a memory block corresponding to the request message through the virtual machine corresponding to the function number that corresponds to the service to be processed. 22. The PCIE device according to claim 19 or 20, wherein the obtaining module is specifically configured for: read, according to the address of the BAT corresponding to the service to be processed, the BAT corresponding to the service to be processed; determine, according to the type of service of the service to be processed, a base CLAT entry address corresponding to the type of service of the service to be processed of the BAT corresponding to the service to be processed; determine an entry displacement according to the service number of the service to be processed; read a CLAT entry corresponding to the entry displacement and the CLAT entry base address that corresponds to the service type of the service to be processed; and read operating information in a block of memory corresponding to the CLAt input. 23. The PCIE device according to claim 21, wherein the CLAT is a single level CLAT or a multiple level CLAT; the CLAT of multiple levels comprises N levels of CLAT, where N> 2 and N is an integer number; each level of CLAT comprises multiple entries; each entry in the nth level of CLAT is used to indicate a direction of the level (n + 1) -this CLAT, where 2 <n <n + 1 <N and n is an integer; and each entry in the N-th level of CLAT is used to indicate a first address of one of the memory blocks. 24. The PCIE device according to claim 23, wherein the sizes of the multiple memory blocks assigned to the service to be processed are identical; Y The obtaining module is specifically configured to: when the CLAT is a CLAT of a single level, multiply the service number of the service to be processed by a size of a memory occupied by the operating information of the service to be processed, and divide the result of the multiplication by the size of the block of memory assigned to the service to process, to obtain a quotient and a remainder for the CLAT of a single level, where when the rest for the CLAT of a single level is greater than 0, the entry displacement is equal to the quotient for the CLAT of a single level, or when the rest for the CLAT of a single level is equal to 0, the entry displacement is equal to the quotient for the CLAT of a single level minus 1; or when the CLAT is a CLAT of multiple levels, where the entry displacement comprises an entry displacement of each level of CLAT, calculate an entry displacement of the X-th level of CLAT in the following way:

where when a remainder for the x-th level of CLAT is greater than 0, the entry offset of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT; or when a remainder for the x-th level of CLAT is equal to 0, the entry displacement of the x-th level of CLAT is equal to a quotient for the x-th level of CLAT minus 1, where A is the size of a memory occupied by the operation information of the service to be processed, B is the size of the memory block assigned to the service to be processed and C is the number of entries in the N-th level of CLAT.